msr imitates kragen-discuss

Dave Long dave.long at bluewin.ch
Thu Oct 19 13:31:46 EDT 2006


That didn't take so long...

"Re: color holograms with macroscopic technology"
<http://lists.canonical.org/pipermail/kragen-discuss/2002-June/ 
000833.html>
> which leads me to wonder -- could one
> use the same technique instead of a
> heap?  Modules could spread state in
> an address space; if there were any
> interference, they could re-place it
> (a spatial backoff, like Ethernet's
> temporal one).

"Samurai - Protecting Critical Heap Data in Unsafe Languages"
http://lambda-the-ultimate.org/node/1730#comment-21630
> Previous approaches to eliminating these errors attempt to eliminate  
> all unsafe memory operations in a program. We present Samurai, a  
> runtime system that allows programmers to selectively identify heap  
> objects that are critical to correct execution of their program.  
> Samurai supports operations to consistently read and update critical  
> data and probabilistically guarantees that no other memory updates in  
> the program will corrupt critical data. Samurai uses replication and  
> randomization to provide these consistency guarantees.

...although it looks like they still lean on the crutch of an  
allocator, rather than relying on randomization and using the back-off  
scheme above.

-Dave



More information about the Kragen-discuss mailing list